Digital Twins in Manufacturing: Separating Hype from Reality
In this white paper
- Defining digital twins
- Origins of digital twins in manufacturing
- Categorizing digital twin applications
- Types of digital twin applications in manufacturing
- The business value of digital twins
- Common implementation challenges
- Best practices for implementing digital twins in manufacturing
- Conclusion
- Download
As manufacturers continue to pursue the benefits of digital transformation through Industry 4.0 initiatives, there is much noise to overcome on the road to success.
WWT firmly believes that integrating digital technologies into production facilities and operations has significant upside, as manufacturers can expect safer, faster, simpler, cheaper and more sustainable outcomes. We also appreciate that successfully implementing advanced technologies requires a certain amount of strategic planning. It can be incredibly disappointing, after all, when a hyped investment fails to deliver results on the factory floor.
Despite the industry's historical reluctance to get ahead of digital transformation, further delaying investment in Industry 4.0 initiatives can be costly. It can even lead to lost market share as competitors leapfrog each other to claim the rewards of early adoption.
Manufacturers trust WWT's industry experts to see past the hype on a variety of IT solutions touted to deliver on the Industry 4.0 narrative. At the top of this list is digital twin technology, which promises a significant step forward in how products are designed, manufactured, serviced and operated across the industry.
This white paper can help separate the hype from reality regarding the use of digital twin technology in manufacturing.
Defining digital twins
Several definitions of the term "digital twin" exist today, each with subtle variations. This lack of agreement on a definition can cause confusion about what a digital twin really is, what it can do and the overall scope of the technology.
Before we present our preferred definition, it's useful to explore the root of this confusion.
The main sticking point involves whether to include the concept of "simulation" in the definition of digital twin. Many believe simulation should be excluded from the category because (a) simulation tools existed before the advent of Industry 4.0 as a concept, and (b) the term does not fully capture or represent the true advanced capabilities of digital twins in terms of technical innovation.
Completely excluding simulation from the definition of digital twin is difficult to support for two reasons.
- First, like simulation tools, many other elements of digital twin technology existed before "Industry 4.0." was coined. Even the term "digital twin" itself predates the arrival of Industry 4.0.
- Second, the ability to test multiple scenarios before making physical changes to the system (e.g., improving system design, error avoidance, process optimization, cost savings, etc.) is often touted as a key benefit of digital twin technology. Like it or not, this ability describes simulation — which is why we believe the term rightfully plays a role in the term's full definition.
How we define "digital twin"
Our preferred definition, in the context of the manufacturing industry, is as follows:
- A digital twin is a digital representation of a physical object, system, process, entity or some combination thereof;
- It is created using data from sensors, Internet of Things (IoT) devices, Industrial Internet of Things (IIoT) devices including Industrial Control Systems (ICS) and other sources to mirror the real-world counterpart in the digital space; and
- It allows for the monitoring, analysis, simulation and/or remote control of the physical entity, either in real-time or historically.
Origins of digital twins in manufacturing
The first use of digital twin technology originated in 1960 when NASA orchestrated a rescue mission to safely return Apollo 13 to Earth after an oxygen tank explosion damaged the spacecraft's engines. NASA used multiple digital simulators to create a living model of the Apollo mission, including the harsh environment of space, to help identify the root cause of the explosion and explore possible countermeasures to prevent similar failures in the future.
Roughly 40 years later, Michael Grieves is credited with first applying the concept of digital twins to manufacturing in the form of a conceptual model underlying the formation of a Product Lifecycle Management (PLM) center.
Industrial control systems
Considering NASA's early use of digital twins, there is an argument to be made that many manufacturers already have some form of digital twin technology deployed on-site in the form of SCADA (supervisory control and data acquisition) systems or DCSs (distributed control systems). Both industrial control systems graphically and digitally represent process equipment in an industrial facility. They often offer access to near-real-time data that can be used to remotely control processes with the click of a button. However, many suggest that more is needed to fully meet the capabilities of a modern digital twin.
Simulation modeling
Simulation modeling is another common example of digital twin usage in factories and other process plants. Depending on the richness of the data used to contextualize the model, simulation modeling can be highly accurate, both from a process perspective and a 3D modeling perspective. Early simulation modeling software includes Arena by Rockwell Automation in 2000 and later Simio, LLC.
In the past 15 years, simulation modeling technology has advanced significantly. It now forms the basis for what we think of as modern digital twin technology. Many vendors offer their own version of simulation modeling software, and some real-world applications have already delivered powerful process optimization results for manufacturers (e.g., see examples from Rockwell Automation and Simio in the two graphics to the right).
As we approach more advanced applications, it's clear that digital twins can be a complex topic to grasp. Nonetheless, digital twin technology is being used in various forms and levels of maturity today across the manufacturing world — even if much of today's narrative speaks of an advanced future state.
Categorizing digital twin applications
We commonly hear the term digital twin being used by clients and peers in a generic manner. However, it's becoming increasingly evident that digital twin often means something different to different people. Part of this can be explained by the many possible types, functions and lifecycle phases where digital twin technology is applicable.
So, how many applicable use cases are there for digital twins in manufacturing?
In their article, Decoding Digital Twins: Exploring the 6 Main Applications and Their Benefits, IoT Analytics considered this question from an industry-neutral standpoint. By categorizing the seven most common use cases, five hierarchical levels and five lifecycle phases, the article suggests that there are around 210 different combinations of digital twin applications.
IoT Analytics then narrowed down the list to the top six most common applications across all industries:
- Twins for system prediction
- Twins for system simulation
- Twins for asset interoperability
- Twins for maintenance
- Twins for visualization
- Twins for product simulation
Building on IoT Analytics's work, WWT embarked on an effort to refine this research by focusing on the specific implications for manufacturers.
Types of digital twin applications in manufacturing
Scope
As a first step, we categorized the scope of possible digital twin solutions by focus area, ordering them from least to most complex.
- Product twins: These digital representations of individual products or product components enable manufacturers to run simulations before and during the manufacturing process in key areas such as product design, testing and optimization. Outcomes include enhanced product quality, performance, durability and usability.
- Asset twins: These are digital replicas of physical assets (usually machines or resources) within a manufacturing environment. They include all relevant data regarding the operation, condition and maintenance history of the asset in question. By integrating real-time data from sensors and historical performance data, manufacturers can optimize the usage, maintenance and lifecycle management of their assets.
- Process twins: With a focus on driving efficiency and productivity across operations, process twins enable manufacturers to replicate process machines, assets and resources to simulate, analyze and optimize procedures and workflows. Outcomes include streamlined production lines that drive more productivity and efficiency while reducing waste.
- Factory or plant twins: Holistic digital replicas of an entire manufacturing plant or factory that integrate various data sources and systems across the facility. They can model a manufacturing environment's physical layout, machinery, workflows, energy systems and even human interactions. Outcomes include a 360-degree view of factory or plant operations that makes it easier for manufacturers to analyze and optimize operations.
- Supply chain twins: This type of digital twin is an end-to-end representation of a manufacturer's supply chain network, including inventory, logistics and distribution. They can help manufacturers facilitate scenario planning, risk management assessments, and optimization initiatives designed to make supply chains more resilient and efficient.
- Digital thread: As the information backbone for digital twin technologies, digital threads link a manufacturer's disparate systems and processes to provide a holistic view of a product's development, production and operational use. Put another way, digital threads are communication frameworks that enable a connected data flow and integrated view of an asset's data throughout its lifecycle.
Applications
Next, we can categorize digital twin solutions by specific application in manufacturing, again from least to most complex:
- Simulation: Creating a simulation that can be modified as a virtual test bed to prototype new scenarios. This digital twin application relies on historical data to enrich the model and leverages rapid iteration to avoid the time and costs of physical testing.
- Visualization: Digital twins can help manufacturers better visualize performance. Despite this application's limited ability to simulate or control the system, it can accelerate a manufacturer's understanding and improve the quality of decision-making.
- Interoperability: This digital twin application marries the interoperability of people, processes and technology to help manufacturers test combinations, understand challenges and integrations, and plan how best to build interoperability into solutions.
- Prediction: Digital twins can help manufacturers predict results or outcomes before they occur. Classic examples include using digital twins to monitor the condition of equipment and predict when machinery or components might need maintenance or fail — based on real-time input from system or sensory data. This application can minimize downtime as well as maintenance costs.
- Remote system control: This application allows manufacturers to visualize and remotely control a system, often through a centralized control center where engineers and operators can monitor production lines live via the digital twin. Remote monitoring and control have become increasingly important, especially for manufacturing operations that demand precision, are hazardous, or are geographically dispersed.
- AI-driven: The integration of AI with digital twin technology has opened up new possibilities in manufacturing for advanced analytics, predictive maintenance and autonomous decision-making. AI evolutions can enhance the ability of digital twins to learn from disparate data sources, predict future outcomes, and make more accurate real-time decisions.
Systems
Third, we defined the lifecycle phases for a manufacturing system using a digital twin — from the design phase through subsequent Build, Operate, Optimize, Maintain and Decommission phases.
This model outlines a set of possible digital twins in manufacturing by plotting the scope, applications and system phases. For example, we can consider a digital twin for a process used to apply simulation during the Design phase, plotted in the example below.
Using this model and data from existing use cases (both customer and internal experience data), with supplementation from various online publications, we were able to break down the model further to extract insights into the different opportunities for digital twin usage in manufacturing today.
Simulation insights from digital twins
In today's manufacturing industry, the scope of simulation use cases via digital twin technology is vast and relatively mature. In fact, we're able to find digital twin simulations in use in nearly every phase of the lifecycle of a manufacturing and supply chain business. A digital twin used for simulation allows users to model existing scenarios as well as new ones to compare results. The value of the resulting output is dependent on the contextual richness of data used to build the model.
Visualization insights from digital twins
Similarly, the range of potential visualization uses for digital twin technology is also relatively mature, as indicated by the figure below. Physical subjects can be visualized across the entire scope levels, from product to digital thread, and can be utilized to test various scenarios across phases, from Design through Decommission. Such visualizations will be used throughout the lifecycle of the subject, and the best versions allow for change to be incorporated with low input costs and continuous improvement.
Interoperability insights from digital twins
The use of digital twins for interoperability often occurs during the Design phase where twins can be leveraged to test the compatibility of the systems, processes and people that will interact with or enrich the subject. During the Operate phase, there is limited opportunity for the functionality of interoperability from the twin. Similar to the Design phase, the Optimize phase is ripe for interoperability applications of digital twins.
Prediction insights from digital twins
Prediction is a strong use case for digital twins. Various applications can be found to predict the impact of proposed system changes or even changes to day-to-day operation across the Design, Operate, Optimize and Maintain phases. The predictive capacity of digital twin technology relies on the quality and richness of the data set, which directly impacts the accuracy and reliability of the digital twin outputs used to shape business decisions.
Control insights from digital twins
Digital twins that are used to control systems are generally concentrated more at the product, asset and process levels of our scope spectrum, as these tend to be driven by a desire to reduce discrete instances of complexity rather than control larger systems. Those seeking to use digital twins to control larger systems should start by connecting existing product, asset and process twins to create a digital thread. A manufacturer's ability to do this depends on access to technology that enables scalability and/or integration between the different digital twins.
AI-driven insights from digital twins
Finally, AI-driven digital twins leverage the functionality of each application to create insights that allow the AI model to make decisions. In today's manufacturing landscape, AI-driven digital twin applications are in a state of relative infancy; however, some progress is being made in the Operate and Maintain phases.
Across the Scope, Phase and Application tangents in the model above, we can see that as digital twin applications increase in scope and complexity, the number of existing use cases diminishes. Which makes sense given the state of the technology's evolution. However, this landscape will change as digital twin technology matures and the number of projects increases to serve the expected outcomes that manufacturers continue to explore.
The business value of digital twins
Focus on outcomes
The specific problem a manufacturer wants to solve will determine the type of digital twin solution(s) that should be implemented. To properly scope a prospective digital twin solution, manufacturers should make sure to fully consider the expected outcomes and business implications of each option. You should also consider each solution's ability to scale with your business as it grows and how anticipated results may change in the future given your overall IT infrastructure modernization roadmap.
Solution maturity
Due consideration should also be given to the relative maturity of the digital twin solution in question. While solution maturity is often inversely proportional to solution value, this relationship can change over time.
For example, consider a manufacturer that implements a digital twin with the express intent of using it as a simulation model — a use case relatively high in maturity as long as proper consideration is paid to the manufacturer's existing technology and architecture. As the landscape of use cases inevitably matures, the manufacturer might very well adapt its simulation twin to a more complex use case, such as providing remote control over an ICS or even enabling AI-driven outcomes. Understanding the maturity curve of digital twin solutions can help manufacturers achieve higher returns on their investments in the future.
Examples of business results
Some of the typical outcomes we're seeing manufacturers pursue with digital twin technology include:
- Improved operational efficiency: Digital twins can optimize processes, reduce downtime and enhance resource utilization. They can also enable better planning, scheduling and execution while driving productivity gains.
- Predictive maintenance: Digital twins can predict equipment failure by analyzing real-time data, thereby enabling proactive maintenance. This can minimize unplanned downtime, reduce maintenance costs and prolong equipment lifespan.
- Enhanced product quality: Digital twins can monitor production lines in real time, identifying defects and other issues early. This can lead to higher-quality products, reduced waste and fewer rework instances.
- Faster time-to-market: Simulating designs and processes with digital twins can accelerate product development cycles. They allow for rapid prototyping, testing and iteration, leading to faster product launches.
- Cost savings: Save money by leveraging digital twins to optimize processes and workflows, minimize downtime, and reduce waste. Predictive maintenance can also minimize the occurrence and impact of costly equipment failures.
- Improved decision-making: Digital twins can deliver actionable insights based on real-time data analysis. This can promote more informed and intelligent decision-making on key topics like production, maintenance and resource allocation.
- Supply chain optimization: Digital twins can be used to optimize inventory levels, improve logistics and enhance coordination among supply chain partners, leading to better responsiveness to market demands and reduced operational costs.
- Remote monitoring and control: Operators can monitor and manage manufacturing processes remotely through digital twins, enabling quicker interventions and adjustments and reducing the need for physical presence.
- Energy efficiency: Digital twins can help optimize energy consumption by simulating and analyzing energy usage patterns. This can lead to more efficient utilization of resources and lower energy costs.
- Customization and personalization: Digital twins can efficiently customize products to meet individual customer needs. This enables manufacturers to produce more personalized products at scale.
- Lifecycle management: Throughout a product's lifecycle, digital twins can provide valuable insights for continuous improvement, better design iterations and product performance enhancements.
Common implementation challenges
As the list of potential outcomes from digital twins grows, so does the inherent complexity of these useful systems. More complexity likely means more hurdles for manufacturers to overcome on their journey to digital twin success.
We've outlined some of the top challenges manufacturers might encounter here:
Data integration and quality
Digital twin outputs ultimately reflect the quality of the data sources available. They also operate under the assumption that more and better quality data will produce better results. However, gathering and integrating quality data from various sources across the manufacturing ecosystem can be a complex task. Inconsistent data formats, siloed information, and varying states of data quality pose real challenges to creating accurate and reliable digital twins.
Costs and investments
Implementing digital twin technology often requires significant investment in sensors, IoT devices, software, infrastructure and skilled personnel. Despite important promises of return on investment, the capital costs of entry can prevent digital twin projects from getting off the ground in the first place.
Interoperability and standards
It can be challenging to ensure compatibility and interoperability among the different systems, devices and software used within a manufacturing environment. The common lack of standardization and disparate technologies, both in terms of age and functionality, can hinder the seamless integration needed for digital twin solutions.
Complexity of models
Developing comprehensive and accurate digital twin models that reflect the complexities of real-world manufacturing processes or systems can overly complicate intended outcomes. Many start with one intention but quickly become overwhelmed by an array of alternate possibilities. This can result in significant scope creep that impacts budgets and timelines. Navigating digital twin complexity not only requires expertise in modeling, simulation and domain-specific knowledge, but it also requires expertise in project and stakeholder management.
Security and privacy concerns
When implementing a digital twin, it is critical to protect sensitive data and intellectual property while ensuring the security of both physical assets and the digital twin models themselves. Cybersecurity threats and vulnerabilities in interconnected systems pose significant risks.
Skill gap and workforce training
Implementing digital twin technology requires a workforce skilled in data analytics, IoT, simulation and other relevant technologies. That is why bridging the skills gap and providing adequate training to employees is so crucial when it comes to adoption.
Change management and cultural shift
Adopting digital twin technology often necessitates organizational changes and a cultural transformation. Resistance to change, a lack of awareness or a reluctance to embrace new technologies can all impede adoption.
Scalability and complexity
It can be tough to scale digital twin projects across more extensive manufacturing operations or complex systems. Manufacturers should strive to ensure that their digital twins remain effective and scalable as the underlying manufacturing environment evolves.
Regulatory and compliance issues
Compliance with industry regulations and standards, particularly concerning data privacy, intellectual property rights and safety, can pose hurdles to digital twin implementation. Be sure you know what laws and regulations are applicable to your operations and plans for digital twin adoption.
Best practices for implementing digital twins in manufacturing
We've found the following best practices to be crucial in holistically scoping the design, implementation and scaling of digital twin solutions in manufacturing:
- Define clear objectives: Clearly define the goals and objectives you aim to achieve by investing in digital twin technology.
- Start with a pilot project: Begin with a smaller-scale pilot project before scaling up. This helps in testing the feasibility, identifying challenges and validating the effectiveness of your digital twin concept.
- Data strategy and integration: Develop a robust data strategy that addresses data collection, integration, storage and security. Ensure the compatibility and seamless integration of data from all relevant sources across your manufacturing ecosystem.
- Choose the right technology: Select appropriate technologies and tools that align with your use case. Consider scalability, interoperability and the ability to handle real-time data for simulations and analysis.
- Collaboration and cross-functional teams: Foster collaboration between different departments and involve cross-functional teams in the implementation process.
- Invest in skills and training: Provide training and upskilling for employees involved in digital twin implementation. Equip them with the necessary skills to utilize the technology effectively.
- Ensure security and privacy: Implement robust cybersecurity measures to safeguard sensitive data and protect the digital twin infrastructure from potential cyber threats. Prioritize data privacy and compliance within the context of relevant regulations.
- Iterative approach and continuous improvement: Implement digital twins iteratively, allowing for continuous improvement and refinement. Gather feedback, analyze results and make necessary adjustments to enhance effectiveness.
- Scalability and flexibility: Design your digital twins with scalability in mind. Ensure they can accommodate future growth, new technologies and evolving manufacturing requirements without significant rework.
- Vendor and partner selection: Choose reliable vendors or partners who can provide expertise, support and technology solutions for digital twin testing and implementation.
- Create a change management plan: Develop a change management plan to communicate the benefits, address any resistance, and facilitate a smooth transition while anticipating future changes.
- Measure and evaluate performance: Define key performance indicators (KPIs) and metrics to measure the success of your digital twin implementation. Regularly assess performance against these indicators to track progress and identify areas for improvement.
Conclusion
Although "digital twin" is often used as a catch-all for a digital manufacturing replica, there are many different types of digital twin solutions, each serving a different set of desired outcomes.
The challenge will be building a thread that brings all of these potential use cases and outcomes together to realize measurable business value. As more and more manufacturers look to adopt digital twin technology for the promised benefits, it's worth remembering that the scalability and flexibility of a digital twin solution should be considered early on, so manufacturers make informed technology decisions from the outset.
The critical elements of technology that will shape your ability to scale digital twin technology are:
- Your data strategy, architecture and integrity
- The networks used to transport your data
- The maturity of your compute and storage strategies and infrastructure
- The security requirements needed to maintain the integrity of your operations and systems
To truly be successful with digital twins, manufacturers need access to technical expertise as well as a deep understanding of industry trends, requirements and limitations. WWT has built a great advantage in respect thanks to our approach, which couples industry knowledge with domain expertise and versatility by embedding industry experts within all of our technical teams that focus on digital twin technology.
Importantly, our team has more than 30 years' worth of experience building physical twins in our Advanced Technology Center (ATC). Advancing and augmenting that experience with our ability to develop and test digital twin solutions in our new AI Proving Ground provides another tool in our tool belt to help manufacturing clients simulate future projects by creating production-like environments in a way that alleviates the constraints of cost, time and complexity.